[digitalradio] Why even use SS, a waste of resources?
Now lets cut to the chase: THE USE OF SPREADSPECTRUM, THAT IS, THE USE OF BANDWIDTH EXPANSION TECHNIQUES BY ADDING PSEUDORANDOM DATA, NOT CREATED FROM THE USER INPUT INFORMATION DATA, IS OF NO ADVANTAGE IN IMPROVING THE END TO END PERFORMANCE OF A LINK, WHEN COMPARED WITH PROPERLY SELECTED MODERN ENCODING AND MODULATION TECHNIQUES. What I am proposing for consideration is the point that for a given transmission bandwidth, and a given end to end data transmission rate (user information), the bits added should actually perform an error reduction function and interference mitigation function. This can be performed using with tradition FEC codes and in modulation selection and encoding (PSK, MFSK,Multicarrier PSK, M-ARY FSK, multicarrier M-ARY FSK, etc.). My point is, why add bits to the transmission that at the receive end, do not improve the performance. For your consideration of the above, I repeat some previously stated basics: (1)Any proper transmission encoding coding scheme, will, as one of its first steps, scramble or randomize the incoming data, in order to provide a uniformly random data stream to the subsequent steps in the process. These randomizers come in a few well defined, published, forms, so it is not that hard to derandomize the result , once you have demodulated, and stripped off the FEC layers. This is typically the first and last step in an end to end process. This process does not produce any encoding or bandwidth expansion. It is a bit in, a bit out process. (2)FEC coding layers, to combat, frequently with one type of FEC, for low signal to noise ratio (QRN)(white noise), inherent in weak signal work to correct random errors, and then outside (around) of the previous FEC, additional layers of FEC, usually a type appropriate to combat bursty errors of the type caused by the time carrying interference environment typical of QRM and atmospheric QRN. (3)Time diversity coding, to combat the channels dispersive distortion in time over HF (short baud bad, long baud good), and frequency selective, but short duration, fading. Incidentally the short baud bad is one reason why spreading tends to underperform on real HF circuits compared to a flat white noise channel in a laboratory environment. (4)Finally, mapping the encoded transmit data into unique modulation states. This is most commonly done as frequency and phase conditions. For example, frequency diversity, in the form of encoding the source to allow it to be transmitted as adjacent multiple carriers or are single carriers on multiple frequencies, is needed to combat the frequency selective fading present on HF paths and to make use of frequencies that at any given instant (in this case, instant = the symbol time) have less noise (QRM) present. There is a practical limit to what can be done in a single carrier system with encoding on HF circuits in particular, because the dispersive (multipath) nature of the HF path is hash on short baud transmissions (high symbol rate). There are a number of ways to reduce the symbol rate of the actual encoded transmitted bits. Changing from BPSK to QPSK actually creates two orthogonal synchronous BPSK transmissions at half (longer) the symbol rate. (FYI: Changing to OFFSET QPSK results in no symbol rate reduction) Using M-Ary FSK where the number of frequencies in the set and the symbol rate are inversely related. For example. Assume a conventional 50 baud(synchronous) FSK transmission. Each transmit symbol is 20ms long. Changing this directly to 8-ary FSK creates eight distinct frequencies, the particular frequency in this case determined by the value of three bits of transmit data used to encode a single transmit baud, that at are used one at a time, with a symbol that is now 160 ms long or 6.25 baud. The result is the same (longer symbol times, easier HF transmissions) with changing from a single psk carrier to multiple adjacent, simultaneous psk carriers, each carrying part of the FEC encoding data stream. In addition to transmit baud rate reduction (symbol time duration increase), multi frequency systems, both single carrier or multiple carrier, can be used as a time diversity encoding to combat dynamic frequency selective degradations such as QRM from other users, QRN from atmospherics, and fading. One final point that should be obvious by now; SS is not necessary to whiten the noise in the transmission bandwidth. In fact, there are more efficient techniques, described above, to do the same thing, described above. In fact, if your transmission bandwidth has a uniform rate of interference, either QRM or QRN, SS is if no help at all. The only way to improve the channel performance is FEC and other forms of mapping the input user data in a deterministic manner, to best match (compensate for) the impairments of the channel. Lester B Veenstra MØYCM K1YCM mailto:les...@veenstras.com
Re: [digitalradio] Why even use SS, a waste of resources?
Lester, Months of testing of all available modes on a 200 mile, weak signal, path on 432 MHz support what you say. Contestia (or Olivia, but slower) has surfaced as the most reliable mode we have found in the difficult environment of signals marginally above the noise, fading (QSB) as deep at 5 s-units, Doppler shift, and Doppler spreading. ROS's spread spectrum simply fails completely, as do any of the PSK modes. Contestia surpasses Olivia simply because it takes only half the time that Olivia takes to pass information, and for our purposes of ragchewing, the constraints of all upper case are not a problem. If you do not like all upper case, in fldigi we have added an option to use all lower case... 73, Skip KH6TY On 7/14/2010 3:51 AM, Lester Veenstra wrote: Now let's cut to the chase: * * *THE USE OF SPREADSPECTRUM, THAT IS, THE USE OF BANDWIDTH EXPANSION TECHNIQUES BY ADDING PSEUDORANDOM DATA, NOT CREATED FROM THE USER INPUT INFORMATION DATA, IS OF NO ADVANTAGE IN IMPROVING THE END TO END PERFORMANCE OF A LINK, WHEN COMPARED WITH PROPERLY SELECTED MODERN ENCODING AND MODULATION TECHNIQUES.* * * What I am proposing for consideration is the point that for a given transmission bandwidth, and a given end to end data transmission rate (user information), the bits added should actually perform an error reduction function and interference mitigation function. This can be performed using with tradition FEC codes and in modulation selection and encoding (PSK, MFSK,Multicarrier PSK, M-ARY FSK, multicarrier M-ARY FSK, etc.). My point is, why add bits to the transmission that at the receive end, do not improve the performance. For your consideration of the above, I repeat some previously stated basics: (1) Any proper transmission encoding coding scheme, will, as one of its first steps, scramble or randomize the incoming data, in order to provide a uniformly random data stream to the subsequent steps in the process. These randomizers come in a few well defined, published, forms, so it is not that hard to derandomize the result , once you have demodulated, and stripped off the FEC layers. This is typically the first and last step in an end to end process. This process does not produce any encoding or bandwidth expansion. It is a bit in, a bit out process. (2) FEC coding layers, to combat, frequently with one type of FEC, for low signal to noise ratio (QRN)(white noise), inherent in weak signal work to correct random errors, and then outside (around) of the previous FEC, additional layers of FEC, usually a type appropriate to combat bursty errors of the type caused by the time carrying interference environment typical of QRM and atmospheric QRN. (3) Time diversity coding, to combat the channels dispersive distortion in time over HF (short baud bad, long baud good), and frequency selective, but short duration, fading. Incidentally the short baud bad is one reason why spreading tends to underperform on real HF circuits compared to a flat white noise channel in a laboratory environment. (4) Finally, mapping the encoded transmit data into unique modulation states. This is most commonly done as frequency and phase conditions. For example, frequency diversity, in the form of encoding the source to allow it to be transmitted as adjacent multiple carriers or are single carriers on multiple frequencies, is needed to combat the frequency selective fading present on HF paths and to make use of frequencies that at any given instant (in this case, instant = the symbol time) have less noise (QRM) present. There is a practical limit to what can be done in a single carrier system with encoding on HF circuits in particular, because the dispersive (multipath) nature of the HF path is hash on short baud transmissions (high symbol rate). There are a number of ways to reduce the symbol rate of the actual encoded transmitted bits. Changing from BPSK to QPSK actually creates two orthogonal synchronous BPSK transmissions at half (longer) the symbol rate. (FYI: Changing to OFFSET QPSK results in no symbol rate reduction) Using M-Ary FSK where the number of frequencies in the set and the symbol rate are inversely related. For example. Assume a conventional 50 baud(synchronous) FSK transmission. Each transmit symbol is 20ms long. Changing this directly to 8-ary FSK creates eight distinct frequencies, the particular frequency in this case determined by the value of three bits of transmit data used to encode a single transmit baud, that at are used one at a time, with a symbol that is now 160 ms long or 6.25 baud. The result is the same (longer symbol times, easier HF transmissions) with changing from a single psk carrier to multiple adjacent, simultaneous psk carriers, each carrying part of the FEC encoding data stream. In addition to transmit baud rate reduction (symbol time duration increase), multi frequency systems, both single
